Culture-independent analysis of fecal microbiota in infants, with special reference to Bifidobacterium species


  • Edited by W. Kneifel

*Corresponding author. Tel.: +81 48 467 9562; fax: +81 48 462 4619, E-mail address:


Fecal microbiota of 31 breast-fed, 26 mix-fed, and 11 bottle-fed infants were analyzed by using terminal restriction fragment length polymorphism (T-RFLP), and culture method. We first determined the total and cultivated bacterial counts in infant fecal microbiota. Only approximately 30% of bacteria present in fecal microbiota were cultivable while the remainder was yet-to-be cultured bacteria. Sixty-eight fecal samples were divided into two clusters (I and II) by T-RFLP analysis, and then subdivided into five subclusters (Ia, Ib, IIa, IIb and IIc). There was no clear relationship between clusters and feeding method. A proportion of bifidobacteria was detected in the fecal material by PCR method using species-specific primers. The predominant Bifidobacterium spp. was Bifidobacterium longum longum type (43 samples (63.2%)), followed by B. longum infantis type (23 samples (33.8%)) and B. breve (16 samples (23.5%)). The distribution of Bifidobacterium spp. was similar in the three feeding groups. In contrast, the high incidence of B. breve in cluster I, especially subcluster Ia and B. longum longum type in cluster II, especially subcluster IIa and IIc were characterized by T-RFLP method. Our results showed that the colonization of Bifidobacterium spp. in infant feces correlated with the T-RFLP clusters.


The development of fecal microbiota in infants is important because it contributes to general health by interaction with the developing immune system [1,2]. In addition, it is well known that breast-feeding has specific effects on the appearance of the composition of fecal microbiota compared with bottle-feeding [3–7]. In healthy infants breast-feeding induces the development of a microbiota rich in Bifidobacterium spp. [4,7,8]. In contrast, colonization of other anaerobes in addition to bifidobacteria and by facultatively anaerobic bacteria is often described in bottle-fed infants [4,5,7,8]. The bifidobacteria predominance in breast-fed infants appears to be among the most significant and seems to provide protection against enteric as well as systemic disorders caused by bacterial pathogens [9].

The intestinal microbiota exceed 1 × 1011/g of human feces and is a complex ecosystem [10,11]. More than 99% of the cultivable fecal microbiota is represented by 30–40 known bacterial species. However, recent studies using the 16S rRNA gene clone library revealed that approximately 75% of the clones from human feces were novel phylotypes [12,13]. Furthermore, these novel phylotypes were found in infant feces [14,15].

The terminal restriction fragment length polymorphism (T-RFLP) technique is currently used as a diagnostic and screening method based on its high sensitivity and ability to rapidly acquire precise data [16–19]. By choosing the appropriate number and types of restriction enzymes, we can increase the probability that the resulting arrays of T-RF size distributions more accurately reflect the natural diversity of microbial populations within a sampled community [17,19].

In this study, we investigated fecal microbiota and composition of Bifidobacterium spp. in breast-fed, mix-fed and bottle-fed infants of approximately 30 days of age by using culture method, T-RFLP and PCR, and compared the results of the three dietary groups with respect to fecal microbiota and Bifidobacterium spp.


2.1Fecal samples

Informed consent was obtained from the parents of each infant. The study material was 68 fecal samples taken from 68 infants of approximately 30 days of age, between February and August 2002 at Toshima Metropolitan General Hospital Tokyo, Japan. All infants made a physical checkup at that time. In this study, the infants were classified into three groups based on the feeding method. The first group of infants was breast-fed only (n= 31). The mix-fed group consisted of infants who received breast milk and formula milk (n= 27). The third group of infants was bottle-fed only (n= 11). All infants were healthy full-term infants with no evidence of disease at the first one month of life and had no need for antibiotics and all were born by vaginal delivery. Immediately after culture, the collected fecal specimens were frozen at −80 °C until DNA isolation.

2.2Bacterial count and bacterial cultivation

The sample for bacterial identification was serially diluted from 10−1 to 10−8 with an anaerobic dilution. For total bacterial count, an appropriate dilution was counted microscopically in a Petroff–Hausser bacterial counting chamber. The total counts were expressed as number of bacteria per 1 g wet weight of fecal content. Two other 0.5 g samples of appropriate dilution were spread on EG agar and BL agar. All visible colonies were counted two days after incubation at 37 °C with 100% CO2.

2.3Bacterial DNA extraction from feces

Bacterial DNA was extracted from 100 mg of fecal specimen using an Ultraclean Soil DNA kit (MO BIO, Solana Beach, CA) and FastPrep™ instrument (Bio 101, Vista, CA) according to Clement et al. [20].

2.4PCR amplification

The primers used for PCR amplification of 16S rRNA gene sequences were 27F and 1492R [18]. 27F was end-labeled with the 6′-carboxyfluorescein (6-FAM), which was synthesized by Applied Biosystems Japan (Tokyo, Japan). The amplification reaction was performed in a total volume of 50 μl containing 0.5 μl of extracted DNA, 1.25 U of Takara Ex Taq (TaKaRa Shuzo, Shiga, Japan), 5 μl of 10Ex Taq buffer, 4 μl of dNTP mixture (2.5 mM each), and 10 pmol of each primer. 16S rRNA genes were amplified in a Biometra Thermocycler T gradient (Biometra, Gottingen, Germany) using the following program: 95 °C for 3 min, followed by 30 cycles comprising of 95 °C for 30 s, 50 °C for 30 s, 72 °C for 90 s, and a final extension for 10 min at 72 °C. PCR products of 50 μl were purified by Ultraclean PCR Clean-up kit (MO BIO) and eluted with 20 μl sterile distilled water. Direct detection of bifidobacteria was performed using the method described by Matsuki et al. [21].

2.5T-RFLP analysis

T-RFLP analysis was performed using the method of Sakamoto et al. [18]. The purified PCR product (2 μl) was digested with 20 U of either HhaI or MspI (TaKaRa Shuzo) in a total volume of 10 μl at 37 °C for 3 h. The reaction digest product (1 μl) was mixed with 12 μl of deionized formamide and 1 μl of DNA fragment length standard. The standard size marker was a 1:1 mixture of the size standards GS 500 ROX (Applied Biosystems). Each sample was denatured at 95 °C for 2 min and then immediately placed on ice. The length of the terminal restriction fragment (T-RF) was determined on an ABI PRISM 310 genetic analyzer (Applied Biosystems) in GeneScan mode. Fragment size was estimated by the Local Southern Methods in GeneScan 3.1 software (Applied Biosystems).

2.6Comparisons of T-RFLP pattern profiles

T-RFLPs using restriction enzymes HhaI and MspI were combined and analyzed by BioNumerics version 2.5 software (Applied Maths, Sint-Martens-Latem, Belgium) and compared by Jaccard similarity coefficient analysis and Ward's algorithm.

2.7Statistical analysis

All data were expressed as mean ± SD. Differences between groups were examined for statistical significance using the Student's t-test. A p value less than 0.05 denoted the presence of a statistically significant difference. StatView software (Abacus Concept, Inc., Berkeley, CA) was used for all statistical tests.


3.1Microscopic and cultivated bacterial counts

Table 1 shows the microscopic and cultivated bacterial counts of each group. The total bacteria count of all 68 samples was 10.82 ± 0.59 log10 CFU/g, and cultivated bacterial count was 10.31 ± 0.36 log10 CFU/g. The total and cultivated bacterial counts in breast-fed, mix-fed and bottle-fed infants were not significantly different. The proportion of cultivated bacteria within the total bacteria for all samples was 30.0 ± 19.6%.

Table 1.  Proportions of cultivated bacterial count to total bacterial count in each feeding group
Feeding methodTotal bacterial countaCultivated bacterial countaCultivated/total bacterial count (%)
  1. aMean log/g wet feces ± SD.

Breast feeding10.81 ± 0.6210.25 ± 0.2027.2 ± 17.0
Mixed feeding10.79 ± 0.5610.30 ± 0.4331.7 ± 21.4
Bottle feeding10.89 ± 0.3010.39 ± 0.4630.1 ± 22.3
Total10.82 ± 0.5910.31 ± 0.3630.0 ± 19.6

3.2Distribution of Bifidobacterium spp. by feeding method

Table 2 shows the distribution of Bifidobacterium spp. in 68 infants. The Bifidobacterium longum longum type was the most common taxon (detected in 43 (63.2%) samples), followed by B. longum infantis type (23 (33.8%) samples) and B. breve (16 (23.5%) samples). On the other hand, the B. catenulatum group (10 (14.7%) samples), B. adolescentis (8 (11.8%) samples) and B. bifidum (7 (10.3%) samples) were the subdominant species. There was no significant difference in the distribution of Bifidobacterium spp. among the different feeding groups.

Table 2.  Distribution of Bifidobacterium species in each feeding group
Feeding methodNo. of Bifidobacterium species positive (%)Total
B. adolescentisB. bifidumB. breveB. catenulatum groupB. longum infantis typeB. longum Longum type  
Breast feeding1 (3.2)3 (9.7)8 (25.8)3 (9.7)10 (32.3)17 (54.8)31 (100.0)
Mixed feeding5 (19.2)2 (7.7)4 (15.4)5 (19.2)8 (30.8)17 (65.4)26 (100.0)
Bottle feeding2 (18.2)2 (18.2)4 (36.4)2 (18.2)5 (45.5)9 (81.8)11 (100.0)
Total8 (11.8)7 (10.3)16 (23.5)10 (14.7)23 (33.8)43 (63.2)68 (100.0)

3.3Cluster analysis of fecal microbiota in infants by T-RFLP patterns

Fig. 1 shows the relatedness between the 68 fecal samples analyzed by combination of T-RFLPs derived from HhaI and MspI and Table 3 summarizes the cluster distribution in each feeding group. The T-RFLP pattern of fecal samples showed two clusters (Cluster I (34 samples) and II (34 samples)), and the clusters were subdivided into five subclusters (subcluster Ia (24 samples), Ib (10 samples), IIa (10 samples), IIb (13 samples) and IIc (11 samples)). Comparison of clusters I and II showed no significant difference in the cluster distribution among the feeding groups, although the proportion of breast-fed infant in subcluster Ia was higher than that in other subclusters. The proportions of mix-fed and bottle-fed infant in each subcluster were not significantly difference except for bottle-fed samples in subcluster IIa.

Figure 1.

Dendrogram obtained from T-RFLP patterns, and distribution of Bifidobacterium species in different feeding groups. Open squares: breast-fed infants, closed circles: mix-fed infants, open triangles: bottle-fed infants. Open circles: samples in which PCR detected Bifidobacterium species. aBifidobacterium species detected using species-specific primer.

Table 3.  Distribution of cluster types according to feeding groups
Feeding methodClusterSubclusterTotal
  1. ap < 0.05, compared with other breast-fed infants.

Breast feeding18 (58.1)13 (41.9)15 (48.4)a3 (9.7)5 (16.1)4 (12.9)4 (12.9)31 (100.0)
Mixed feeding11 (42.3)15 (57.7)7 (26.9)4 (15.4)5 (19.2)7 (26.9)3 (11.5)26 (100.0)
Bottle feeding5 (45.5)6 (54.5)2 (18.2)3 (27.3)0 (0.0)2 (18.2)4 (36.4)11 (100.0)

3.4Distributions of Bifidobacterium spp. in T-RFLP clusters

Table 4 shows the detected Bifidobacterium spp. in each cluster. The proportion of B. breve in cluster I was significantly higher than in cluster II, and the proportion of B. breve in subcluster Ia was higher than in other subclusters. Moreover, the proportion of B. longum longum type in cluster II was significantly higher than in cluster I, and those in subcluster IIa and IIc were higher than other subclusters. In contract B. adolescentis in subcluster IIa, B. bifidum in subcluster Ia, IIa and IIc, B. breve in subculster IIa and the B. catenulatum group in subcluster Ib were not detected.

Table 4.  Distribution of Bifidobacterium species in each cluster
ClusterNo. of Bifidobacterium species positive (%)Total
B. adolescentisB. bifidumB. breveB. catenulatum groupB. longum infantis typeB. longum longum type  
  1. ap < 0.05. Comparison of clusters I and II.

  2. bp < 0.05. Comparison of subcluster Ia with other subclusters.

  3. cp < 0.05. Comparison with subclusters Ia, Ib and IIb.

I3 (8.8)3 (8.8)13 (38.2)a4 (11.8)11 (32.4)16 (47.1)34 (100.0)
II5 (14.7)3 (8.8)3 (8.8)5 (14.7)12 (35.3)28 (82.4)a34 (100.0)
Ia1 (4.2)4 (16.7)11 (45.8)b5 (20.8)8 (33.3)12 (50.0)24 (100.0)
Ib2 (20.0)0 (0.0)2 (20.0)0 (0.0)3 (30.0)4 (40.0)10 (100.0)
IIa0 (0.0)0 (0.0)0 (0.0)1 (10.0)2 (20.0)9 (90.0)c10 (100.0)
IIb4 (30.8)3 (23.1)2 (15.4)2 (15.4)6 (46.2)8 (53.8)13 (100.0)
IIc1 (9.1)0 (0.0)1 (9.1)2 (18.2)4 (36.4)10 (90.9)c11 (100.0)


Fecal microbiota has been analyzed in infants by culture methods and differences in fecal microbiota in breast-fed and bottle-fed infants have been described previously [3,5,8]. However, the existence of yet-to-be cultured bacteria in human fecal microbiota was documented by molecular techniques [12,13]. Furthermore, Favier et al. [14,15] also showed that yet-to-be cultured bacteria colonized in the intestinal microbiota of infants. In this study, our results showed that approximately 30% of bacteria present in the fecal microbiota were cultivable. Hence, the culture method is limited in assessing the diversity of the whole infant fecal microbiota. In this study, we used the T-RFLP analysis to compare the intestinal microbiota of healthy breast-fed, mix-fed and bottle-fed infants.

The feeding method is an important determinant of fecal microbiota composition in healthy infants and interaction between the type of milk and fecal microbiota has been demonstrated by using traditional culture methods [5,6,22,23]. However, we noted that the relationship between the type of milk (breast versus formula milk) and T-RFLP cluster (bacterial community) was not clear. One reason for the disagreement between the culture method and T-RFLP analysis was considered to be associated with yet-to-be cultured bacteria. Although the culture method can essentially identifies only cultivable bacterial groups, the molecular methods can detect obligate anaerobic bacteria, which need unique and complex nutrition and strict anaerobic conditions for their growth [10–12]. Furthermore, 45–56% of clones in fecal microbiota in infants determined by DGGE were not identified previously in the human intestinal microbiota [14,15]. The diversity of yet-to-be cultured bacterial community is greater than that of cultivable bacterial community [12,24]. Hence, our results obtained from T-RFLP analysis would be influenced by the diversity of yet-to-be cultured bacteria. However, the influence of feeding method on yet-to-be cultured bacteria is still unclear and there is a need to study the relationship between milk type and fecal microbiota.

Several studies evaluated T-RFLP pattern by cluster analysis and similarity index and these analyses showed the relationship between bacterial community and various environmental, nutritional and diet-related factors [18,25–28]. For example, Schwiertz et al. [25] compared the similarity index of DGGE band patterns in preterm infants and showed that fecal microbiota in preterm infants were influenced by environment. In addition, Sait et al. [29] measured the effect of secretory immunoglobulin (sIg) on the intestinal microbiota of 10-week-old mice by using cluster analysis and similarity of T-RFLP pattern and demonstrated that sIg did not play a role in modulating the composition of the bacterial community capable of colonization in the mouse ileum. Thus, molecular techniques can accurately estimate those factors that could influence fecal microbiota.

Although the relation between type of milk and cluster was not clear, our results showed that the distributions of Bifidobacterium spp. (B. breve and B. longum longum type) were associated with T-RFLP clusters. Hence, T-RFLP could reveal that the composition of bacterial community including yet-to-be cultured bacteria influenced the colonization of Bifidobacterium in the infant intestine.

The Bifidobacterium spp. that had been frequently noted in breast-fed infants was B. breve, B. bifidum and B. longum infantis type [3,5,30,31]. In addition, Matsuki et al. [21] also showed that B. breve was the most commonly detected species in infant by using the PCR method. Our results, however, showed that the most common species in breast-fed infants was B. longum longum type but not B. breve. These results point to differences in the distribution of Bifidobacterium spp. in breast-fed infants by investigators [3,5,30], which could reflect differences in breast milk composition, as well as other factors such as geographical differences [31], mother's vaginal microbiota [30], and hospital environment [22].

Little is known about the microbial diversity of the colonic ecosystem in infants and whether the microbiota composition may influence host–microbiota interactions and allergic disease [32,33]. Differences in the neonatal enteric microbial patterns were recently reported to be associated with the development of atopic disorders. Ouwehand et al. [33] and He et al. [32] demonstrated that breast-fed allergic infants harbored an adult-type bifidus flora (B. adolescentis) with reduced adhesive properties to the intestinal mucus. In contrast, our results showed colonization of typical infant bifidobacterial flora consisting mainly of B. bifidum, B. longum infantis type and B. breve in healthy breast-fed infants. It is important to realize that adhesion to the intestinal mucosa is not a prerequisite for colonization by microorganisms and that only adhesive strains are thought to play a key role in immune-modulation. Our study did not discuss the relationship between fecal microbiota and allergic disease. However, if a relationship between reduced population of Bifidobacterium and these diseases could be found, T-RFLP analysis will be useful diagnostic tool of these risk factors. Further work is needed to clarify the association among infant disease, Bifidobacterium spp. and yet-to-be cultured bacteria.


We thank the staff of Toshima Hospital and the families who provided fecal samples for this study.